The Case Against CAS

Content addressable storage (CAS) is getting a reputation and it's not one to be envied.

Over the last couple of years, a handful of vendors led by EMC have launched products that create searchable, object-oriented repositories of fixed-content data. In markets like healthcare, where medical images like scans must be saved indefinitely, CAS seems a key solution.

Yet even as CAS has proliferated with the launch of EMC's Centera in 2002 and the advent of startups like Archivas, Bycast (which OEMs its medical CAS system to HP and IBM), Nexsan, and Permabit — some are crying foul. (See CAS Conundrum.) CAS, they say, is just another invitation to vendor lock-in.

There are several complaints: Most CAS systems don't run on heterogeneous hardware. Once you buy from one vendor, you can't add another. CAS isn't as open as it should be. It's still tough, for instance, to get a system that creates policies for moving data from one area of storage to another without APIs. Tape storage isn't always supported. Retrieving data can be iffy and performance is an issue.

Indeed, CAS has been controversial enough that some vendors have resisted the term in their marketing. (See Archivas Seeks Archiving Action.) Still, the negative publicity surrounding CAS has also encouraged counterclaims.Now Archivas, while not embracing the CAS term, claims hardware independence for its software, in spite of reseller arrangements with Source Code Corporation, Dell, and possibly Hitachi Data Systems. (See Hitachi Heads for the Archives.) Permabit claims to be on the verge of going "software only" in its market approach. A stealth-mode startup called Caringo (no Website yet) says it's readying software to rewrite the story of CAS. And at least one SNIA standards exercise, eXtensible-Access Method Interface (XAM-I), is aimed at interoperability among CAS systems.

Heartening news to be sure. Yet it's difficult not to look at the track record for CAS and wonder if any of this is for real.

Consider the following: There's been little or no effort on the part of EMC and several CAS startups to work on hardware other than their own or their partners' hardware. Indeed, it seems hardware accounts for at least 40 percent of a CAS system's price. (See CAS at a Crossroads.) Little wonder that despite claims of hardware agnosticism by a select few, those vendors encourage lucrative reseller partnerships.

Perhaps we're on the wrong track. As in other areas of storage, "software only" may not be the best approach for CAS. Is heterogeneous hardware compatibility worth the struggle — especially if hardware adds reliability, security, and even flexibility?

There are no easy answers. But there are more announcements on the way. These include hardware expected from HDS (as noted above) and a software announcement from Symantec. That company expects to announce within the next few months a new release of the CAS product Veritas acquired with its purchase of DataCenter Technologies in 2005. (See Veritas Archives Another Startup.)Bottom line? CAS may have earned its rep, but don't expect vendors to repent in an expected fashion. There may be more than one solution. And even as lip service to multivendor hardware and XAM-I proliferates, user pressure seems to be forcing real change. Whether that takes the form of software, hardware, or a combination remains to be seen. Hopefully, it will lead to something better.

— Mary Jander, Site Editor, Byte and Switch

Organizations mentioned in this article: